re.
Once we get to the point where we're feature complete we can and
should think
about API stability. At least with inventory we're not there by a long shot so
I am not particularly willing to trade API clarity for "stability" of sub-par
APIs that we produce and change as we go along.
I'm not talking about changes for changes' sake but about substantial changes
to the model that evolves as we iron out our usecases and workflows.
While I 100% agree with you that we should communicate the changes and provide
migration guide (if not automatic forwarding to new APIs where possible), I
personally don't think we're at the point in the project's life where we can
afford that burden. I personally don't have time to do that that wouldn't be
better spent by burning some of the ever growing backlog of stuff to be done
on inventory so that it can become at least partially usable by clients.
+1, we need to take advantage of this window where back compat is not
overly required and use cases continue to become solidified. Of course
don't go ten steps back to go one step forward. As with everything,
make good trade-offs and don't hang out to dry the folks using your stuff.